"The Dark Side of AI Therapists" - Latest Stanford Research Unveils "Dangerous Biases"

On July 13, Stanford University announced that it had analyzed five AI therapy chatbots and identified instances of "biased expressions" and "failures in crisis intervention." The study revealed discriminatory responses towards schizophrenia and alcohol dependency, as well as inappropriate replies to suggestions of suicide, warning that "scaling up models does not enhance safety." This report was disseminated by TechCrunch, sparking a debate on social media between those calling for "urgent regulation" and others advocating "not to compromise accessibility." Experts see potential in supplementary use cases but emphasize the need to urgently establish safety standards and audit systems.